using big data, ml, graph data base for better citizen...
TRANSCRIPT
SAMAGRA VEDIKA --TELANGANA’S INTEGRATED PLATFORM
Using Big data, ML, Graph data base
FOR BETTER CITIZEN SERVICE DELIVERY AND TRANSPARENCY, ACCOUNTABLE AND EFFICIENT GOVERNANCE
I TE&C DEPARTMENT G OVERNMENT OF TELANG ANA
Student scholarships
Few of the welfare schemes of Telangana Total budget is more than 35,000 Cr
Ration Cards
AasaraPensions
Most of the welfare schemes have eligibility in terms of Income
Raithu Bandhu
Identity Fraud
Quantity Fraud
Eligibility Fraud
These leakages have been controlled
through use of Aadhar & ePOS
Right Beneficiary
Limited means to establish
identify of right beneficiaries.
Research on mitigation is currently
ongoing
No solution in the country as on date
as it requires data of other
departments
Bogus Beneficiary
Non-existent real beneficiary
Duplicate Beneficiary
Multiple registrations by same beneficiary.
1
2
3
Illegitimate Claims Claiming bills
in MNREGA without work
Disproportionate Quantity
Availing more quantity of PDS
Food grains than eligibility
Is the person truly eligible? Limited
means to correctly establish the
eligibility
possible leakages in welfare programs and some still unplugged.
Resolution throughTypes of Fraud Leakage on account of
This is probably as big
as a budget of a minor
department!
Quantum Of Leakages Due To Wrong Inclusion
Total Budget For
2019-20 For
Pensions
₹ 10,000 Cr
Value of just 1% Leakage…
₹ 100 Crores
Eligibility for many benefits is prescribed.
5
People fulfilling one or more of the following conditions listed below shall not be eligible for
Aasara Pension:
Classification Aspect
Self-Economic
Indicators
Having land more than 3.0 acres wet/ irrigated dry or 7 5 acres dry.
Having large business Enterprise (oil/rice mills, pumps, shop owners etc.).
Owners of light and/or heavy automobiles (four wheelers and big vehicles)
Family Based
Parameters
Having children who are Government/Public sector/ Private sector
employment / Out-sourced/Contract.
Having children who are Doctors, Contractors, Professionals and Self
employed.
Government
PensionerAlready receiving Government pensions or freedom fighter pensions.
Others
Any other criterion in which the verification officer may asses by the
manner of lifestyle, occupation and possession of assets rendering the
household as ineligible
Traditional process → ineffective implementation
6
Manual Process
No Digital Trail
High Human Discretion
Reliance on
Aadhar alone Almost no accountability of
officials in either error
Inclusion Error
( Benefit given to
ineligible persons)
Exclusion errors
(Denying eligible
persons)
Effects of ignoring these challenges
Opportunity & Challenges to get Consolidated view
1. Data is in Silos
2. No Common ID
3. Integrated view – (SSOT) Single Source of Truth is not
available
1. Most of the Data is in electronic form
Samagra Vedika One View - Objectives
What is the alternative approach Without using Aadhar or any other ID
But getting the same efficacy In view of Legal restrictions on use of Aadhar
Meta data attributes of an entity present in all data sources
• Following meta data information is available in every data source
Unique Personal Details
– Name
– DOB
– Fathers Name ( In Some)
Unique ID* Number
– PAN or
– Passport no or
– Voter ID or
– Driving LicenseContact Details
– Mobile No. ( In some)
– Address (Res)
– Address (Off)
Photograph (
Some data sets) *Any one ID is present
• All records in all data sources have Name, Address.
• Some records also have DoB, Phone Number, Fathers Name, Photo
• Can a combination of these attributes which are already available in every record be used to identify an entity
• With an Accuracy nearer to Aadhar based linkage
• With no manual intervention
3 “V” CHALLENGES
Examples of variations in Name & Father’s Name
Spelling
Abbreviations
Sequence Variation
Addition/ Deletion
Splitting
• N Radha Murali Krishna
• N Radha Muralee Krishna
• N R Murali Krishna
• N R M Krishna
• Murali Krishna N Radha
• N M Radha Krishna
• N Murali Krishna
• Murali Krishna
• Murali Krishna N Radha
• Radhakrishna N M
13
14
Key Performance Metrics
FOLLOWING TECHNOLOGIES ARE USED
Economic Survey 2019 of GOI has praised Samagra Vedika
Accurate targeting of subsidies
Beneficiaries for Old age pensions
Using Samagra Vedika for new sanctions of Aasara Pensions
18
Total new applications
received
Total eligible applications
(as per Samagra Vedika)
Total ineligible applications
(as per Samagra Vedika)
65,693 59,068
6,625 (10.1%)
Value Rs 16 Cr per anum
Eligibility for Aasara Pensions is now 57 years
New applications are being received (expected about 7 to 8 lakh new pensioners )
65,693 new applications approved by the Districts officials after verification
Are sent to SERP for sanction Aasara Pensions
In Aug 2019 SERP requested ITEC to check the eligibility
Through Samagra Vedika platform
Accurate targeting of subsidies
Predictive analytics based identification of beneficiaries for 2 BHK scheme using Big data
Govt of Telangana has a program to provide 2BHK houses to
economically weaker sections
• Started in 2015 by the Govt. Of Telangana to provide 100% subsidized housing to the poor.
• No beneficiary contribution needed – one of its kind in India.
• Construction cost = 7-8 lakhs/house and total cost including land is 15-20 lakhs/house
20
Total Houses constructed
under the scheme
Total applications received
2000
11,681
In one district
Aligned with the objective of implementation of scheme in the entire state, Govt of Telangana is looking to distribute 2BHK houses to
eligible persons.
The significant expenditure by Govt, and high mismatch in number of applicants and available houses has necessitated a very
careful approach towards allotting the houses to applicants.
21
Key observations
Earlier Beneficiaries
of Housing Schemes
Already owing
HousesFinancially well-off
applicants
It is difficult to correctly identify the right
beneficiaries
• basis the information collected in the
application form.
• basis any other information available with
district adminstration
• Manual system
Some of the applicants have received
subsidized housing earlier, but are
reapplying using a family members name or
their name
Certain applicants have submitted low
incomes certificates even though they
are financially well offSome applicants or their family
members already own a house.
Siddipet Dist. Administration had following key observations.
Predictive analytics using big data used following data sets available
with Govt.
22
❑ Name
❑ Fathers name
❑ Address
❑ Aadhar number
❑ Phone number
❑ Photo of the applicant
❑ Minor info.
Electricity connection
Water connection
House and land database
Old age pension schemes
Vehicles database
Ration card database
datasets available Information provided by applicant
Information about family members not provided
in the application
Common databases are matched with the provided info. and they are further analyzed to bring out valuable insights in
the form of applicant categories
The analysis using Samagra Vedika categorized the applicants in four
categories as follows:
23
Categories -> Category 1 Category 2 Category 3 Category 4
Classification Qualify Qualify with verification Consider as low priority Don’t consider
Not financially well off
No housing benefits previously
accepted
No other welfare schemes prior
From Siddipet - SKS
Count (% of Total) 2363 (20.2%) 2678 (22.9%) 2181 (18.7%) 4459 (38.2%)
Pilot at Hyderabad in Aug 2016
In Hyderabad about 1,00,000 cards were removed in Aug 2016
There was some public resistance due to which people were asked to apply again. About 19,000applied as on Dec 16, 14,000 cards were activated again after verification that the property is verysmall or the four wheeler is taken out of loan etc.
Net about 86,000 cards are removed from August 16.
Total subsidy saved is Rs 4.6 Cr every month from Aug 16 onwards.
The mistake of tagging the vehicle/house to a wrong person is less than 5% which shows the efficiency of the application
THANK YOU